Import and Clean Amazon Product Reviews
- Change Log:
- Imported Amazon review data
- Cleaned up column names
- Removed undeeded columns
- Extracted date from date column.
taras_caramels <- read_csv("./taras_caramels.csv") %>%
janitor::clean_names() %>% select(-id,-profile_name,-images)
## Rows: 233 Columns: 8
## -- Column specification --------------------------------------------------------
## Delimiter: ","
## chr (6): id, profileName, text, date, title, images
## dbl (2): rating, helpful
##
## i Use `spec()` to retrieve the full column specification for this data.
## i Specify the column types or set `show_col_types = FALSE` to quiet this message.
taras_caramels$text <- tolower(taras_caramels$text)
# Extract Dates from test column
taras_caramels$Ymd <- lubridate::mdy(taras_caramels$date)
taras_caramels <- taras_caramels %>% select(Ymd,title:helpful) %>%
rename(review = title)
summary(taras_caramels)
## Ymd review rating helpful
## Min. :2018-05-16 Length:233 Min. :1.000 Min. : 0.0000
## 1st Qu.:2019-09-04 Class :character 1st Qu.:3.000 1st Qu.: 0.0000
## Median :2020-09-28 Mode :character Median :5.000 Median : 0.0000
## Mean :2020-06-27 Mean :3.811 Mean : 0.9828
## 3rd Qu.:2021-06-01 3rd Qu.:5.000 3rd Qu.: 0.0000
## Max. :2021-11-19 Max. :5.000 Max. :118.0000
ggplot(taras_caramels) + geom_histogram(aes(x=rating),bins = 9) +
labs(title = "Count of the Number Reviews by Rating")
p1 <-taras_caramels %>% count(Ymd) %>% ggplot() + geom_col(aes(x=Ymd,y=n)) +
labs(title = "Number of Reviews per Day",y="Number of Reviews")
ggplotly(p1)
Analyize Words Using “sentimentr” Package
Commands from Video (see above)
head(sentiment(taras_caramels$review),25)
## element_id sentence_id word_count sentiment
## 1: 1 1 4 -0.25000000
## 2: 1 2 8 0.35355339
## 3: 1 3 5 0.44721360
## 4: 2 1 5 0.22360680
## 5: 3 1 11 0.00000000
## 6: 4 1 10 0.04743416
## 7: 5 1 9 0.61683333
## 8: 6 1 2 0.53033009
## 9: 7 1 3 0.28867513
## 10: 8 1 4 0.30000000
## 11: 9 1 1 1.00000000
## 12: 10 1 5 -0.50311529
## 13: 11 1 6 0.36742346
## 14: 12 1 1 -0.50000000
## 15: 12 2 3 -0.57735027
## 16: 13 1 1 0.00000000
## 17: 13 2 3 0.00000000
## 18: 14 1 1 0.50000000
## 19: 15 1 3 0.43301270
## 20: 16 1 4 -0.50000000
## 21: 17 1 1 0.50000000
## 22: 17 2 1 0.00000000
## 23: 17 3 1 0.00000000
## 24: 18 1 6 0.00000000
## 25: 19 1 6 -0.30618622
## element_id sentence_id word_count sentiment
sentiment_by(taras_caramels$review)
## element_id word_count sd ave_sentiment
## 1: 1 17 0.378408 0.18358900
## 2: 2 5 NA 0.22360680
## 3: 3 11 NA 0.00000000
## 4: 4 10 NA 0.04743416
## 5: 5 9 NA 0.61683333
## ---
## 229: 229 14 NA 0.06681531
## 230: 230 1 NA 0.50000000
## 231: 231 6 NA -0.20412415
## 232: 232 3 NA 0.08660254
## 233: 233 6 NA 0.24494897
Sentence Structure
taras_sentence <- taras_caramels %>% select(review) %>%
get_sentences() %>%
sentiment()
taras_sentence %>% ggplot() + geom_density(aes(x=sentiment)) +
labs(title="Density Plot of Rating Sentiment",
x=" <-- Negative(-) Bad but Plus(+) Good --> ")
taras_sentence %>% ggplot() + geom_histogram(aes(x=sentiment)) +
labs(title="Histogram of Rating Sentiment",
x=" <-- Negative(-) Bad but Plus(+) Good --> ")
## `stat_bin()` using `bins = 30`. Pick better value with `binwidth`.
taras_sentence <- taras_sentence %>%
mutate(plus_minus = if_else(sentiment >0,"Positve","Negative"))
taras_sentence %>% count(plus_minus,sort =T) %>%
ggplot(aes(x=plus_minus,y=n)) + geom_col() +
labs(title = "Negative vs Postive Ratings",
y="Number of Ratings")
Begin TidyText Package Analysis
taras_tokens <- taras_caramels %>% select(review) %>%
unnest_tokens(word, review)
taras_tokens %>% count(word,sort =T ) %>% top_n(25) %>%
ggplot(aes(x=reorder(word,n),y=n)) + geom_col() + coord_flip()
## Selecting by n
data(stop_words)
taras_clean <- taras_tokens %>%
anti_join(stop_words)
## Joining, by = "word"
taras_clean %>% count(word,sort =T ) %>% top_n(25) %>%
ggplot(aes(x=reorder(word,n),y=n)) + geom_col() + coord_flip() +
labs(title ="Count of Top 25 Words",
x="Count of Words","Token Words",y="Word Counts")
## Selecting by n
Construct a Word Cloud
taras_clean %>%
count(word) %>%
with(wordcloud(word, n, max.words = 200))
